The retrieval component in RAG often leverages search engine indexes and ranking algorithms. Therefore, many traditional SEO factors are still directly relevant:
- Continue Keyword Optimization: Even though AI is semantic, the retrieval step may use keywords or their embeddings. Make sure each content page targets a primary topic (keyword cluster). Use that in the title, headings, URL, and naturally throughout. For example, if a page is about “How to secure IoT devices,” ensure those exact words (and synonyms like “secure connected devices”) are present. As we noted earlier, SGE doesn’t always pick only top-3 results, but being on page 1 certainly helps. High organic rankings equate to higher chance of retrieval. Aim to maintain or improve your search rankings for important queries, as a baseline for GEO.
- Build Quality Backlinks: Links remain a major factor in search retrieval. If your content is widely cited/linked, search engines rank it higher, and thus AI is more likely to pull it. During retrieval, algorithms might favor sources with authority. Earning backlinks through outreach, PR, and good content is thus part of GEO. For example, if you have a comprehensive guide that many other sites link to, Bing’s index likely ranks it high and Bing Chat will be inclined to use it in answers.
- Optimize Meta Tags for Context: The meta title and description of your pages can influence snippet selection in search (and indirectly what text AI might see in the index). Write meta descriptions as a concise summary of the page’s answer to a question. While an AI might not directly use the meta description, that meta might appear in search results and thus in the AI’s retrieved context. If the description says “AcmeCo explains 5 key steps to improve network security,” the AI might read that snippet and know what’s in your page even before clicking.
- Leverage Structured Data in Retrieval: Some AI might use search’s Structured Data. For instance, Google’s internal systems might pass along that your page has FAQ schema or a star rating. While not confirmed, it’s plausible that if two sources are otherwise equal, the one with structured markup (indicating clearly what’s on the page) could be favored because it’s easier to trust and extract from. Continue implementing relevant schema as earlier, which aids SEO snippet features and potentially AI retrieval.
- Use Site Search and XML Sitemaps: Ensure your site has an XML sitemap listing all important content and that it’s submitted to search consoles. Also enable IndexNow (a Bing-supported standard for immediate indexing) if your CMS or CDN allows – this can speed up Bing’s awareness of new/updated content, which is useful for Bing Chat freshness. If you have a site search feature, tune it to display SEO-friendly URLs (some site searches have their own results pages that can get indexed). Occasionally, we’ve seen Google index internal search results; while generally not ideal, if an AI were to stumble on that, it could confuse things. So use meta robots noindex on any low-quality internal search or tag pages.
In short, don’t neglect SEO. GEO is additive. Continue doing what keeps you ranking, as those rankings feed the retrieval engines of AI.